|Home | About | Journals | Submit | Contact Us | Français|
Respiratory syncytial virus (RSV; Family Paramyxoviridae, Genus Pneumovirus) is a major respiratory pathogen of infants and children and an emerging pathogen of the elderly. Current management of RSV disease includes monoclonal antibody prophylaxis for infants identified as high risk and supportive care for those with active infection; there is no vaccine, although several are under study. In this manuscript, we review published findings from human autopsy studies, as well as experiments that focus on human clinical samples and mouse models of acute pneumovirus infection that elucidate basic principles of disease pathogenesis. Consideration of these data suggests that the inflammatory responses to RSV and related pneumovirus pathogens can be strong, persistent, and beyond the control of conventional antiviral and anti-inflammatory therapies, and can have profound negative consequences to the host. From this perspective, we consider the case for specific immunomodulatory strategies that may have the potential to alleviate some of the more serious sequelae of this disease.
The human pneumovirus, respiratory syncytial virus (RSV) is the leading cause of lower respiratory tract infection in infants and children throughout the world and is emerging as an important pathogen of the elderly . Nearly all children have experienced a first RSV infection by two years of age, and, although most infections are mild to moderate, approximately 120,000 infants and children per year in the US alone require hospitalization for onset of severe disease. Identified risk factors for severe disease include premature birth, bronchopulmonary dysplasia, immunodeficiency, and congenital heart disease; yet a recent study by Hall and colleagues  documented the fact that most children hospitalized with severe RSV infection were previously healthy, with no identified risk factors, a finding that highlights the problems with strategies that focus on known high-risk children only [Table 1].
Current management for RSV infection includes prophylaxis with humanized monoclonal antibody administered via multiple doses during peak season to aforementioned high risk infants . A vaccine trial in the 1960s concluded with tragic consequences ; although there are currently no approved vaccines for RSV disease, several are currently under study [5, 6]. Treatment of severe disease is supportive care only. Ribavirin, a nucleoside analog that introduces mutations into the RNA viral genome during replication, was previously used routinely for infants hospitalized with RSV. Ribavirin is an effective antiviral, but was determined to have no impact on clinical outcome in otherwise healthy infants . As such, the current American Academy of Pediatrics guidelines do not support the routine use of ribavirin for RSV-infected children . Likewise, glucocorticorticoids, although broadly anti-inflammatory in other settings, result in only minimal clinical impact with respect to the management of acute RSV infection [9, 10].
Thus, there is general agreement regarding the need for better strategies for treating severe RSV disease. Toward this end, there are numerous pharmacologic agents currently in development, most focused on inhibiting virus replication. Among these are inhibitors of virion attachment, inhibitors of membrane fusion, as well as antisense inhibition strategies directed at viral RNA (reviewed in ). Given the poor clinical outcome achieved with ribavirin, which is an agent that is actually very effective at achieving control of virus replication, one wonders whether a purely antiviral approach will be successful. As most of the patients will be seeking clinical intervention only after virus replication has persisted for several days and the associated inflammatory response has led to clinical symptoms, will targeting the virus alone achieve therapeutic success? We will discuss the evidence indicating the need for development of immunomodulatory therapy for the treatment of severe RSV infection.
RSV is classified in the subfamily Pneumovirinae of the virus family Paramyxoviridae; it has a negative-sense linear single stranded RNA genome of >15 kb. The 10 virus genes encode 11 polypeptides, including the non-structural proteins NS1 and NS2, the N nucleoprotein, P phosphoprotein (polymerase cofactor), proteins M, M2-1, and M2-2, the L (large) polymerase, and surface proteins G (attachment), F (fusion) and small hydrophobic (SH). The virion structure consists of a helical nucleocapsid (N protein and the viral RNA genome) enclosed within a lipid bilayer derived from the host cell plasma membrane. To initiate infection, RSV virions interact with glycosaminoglycans on the target cell surface ; the chemokine receptor, CXCR3, cell surface annexin II, and the transit protein nucleolin are among the proteins that have been proposed as specific receptors for RSV binding [13–15]. Entry of the nucleocapsid takes place upon fusion with the target cell membrane. Transcription of viral proteins involves the concerted actions of the virus N, P, L and M2-1 polypeptides; this is followed by genome replication in the cytoplasm via an anti-genome intermediate. Completed infectious virions, which can range from round particles of ~200 nm diameter to long filaments, bud from the target cell surface. Specific details of these events can be found in several recent, excellent book chapters [16, 17].
While one can measure production of proinflammatory mediators from virus-infected primary cells and cell lines in culture, the full impact and pathophysiologic sequelae of the inflammatory response can only be examined in a living organism. In a mammalian model, one can examine the production of proinflammatory mediators, as well as ongoing and progressive outcomes – leukocyte recruitment, pulmonary edema, impaired oxygen exchange, respiratory dysfunction, weight loss, mortality – and likewise outcomes and responses to therapeutic intervention.
There are numerous models of pneumovirus infection available for study; these have been recently reviewed [18–20]. Pneumovirus infection models can be divided into two main categories. Heterologous virus-host models are those in which the human RSV pathogen is used to inoculate non-human mammalian targets. Primate models are formally included in this group (e.g., chimpanzee); their responses to hRSV are most similar to humans. Among the heterologous models that are available for routine laboratory study are the BALB/c and cotton rat models of hRSV challenge. BALB/c mice are inbred, readily available at various ages, and are accompanied by a vast array of sophisticated genetic reagents and gene-deleted strains for detailed, complex study. However, traditional laboratory strains of RSV undergo limited, if any significant replication in bronchial epithelial cells of adult wild-type BALB/c mice, and only in response to very high titer virus inocula. Nonetheless, numerous studies have documented responses to virion inoculation and the impact of gene-deletions on inflammatory pathways and clinical responses, and varying rates of virion clearance have been reported. Furthermore, several groups have recently characterized of clinical RSV isolates with enhanced inflammatory responses [21, 22]. In contrast, cotton rats (Sigmodon hispidus) are semi-permissive hosts for RSV replication, reported as ~100-fold more susceptible to infection than BALB/c mice. In response to a moderate RSV inoculum, replication is observed primarily in the lower airways, in association with mild to moderate bronchiolitis .
In contrast, cognate virus-host models – those in which pneumoviruses related to the human RSV pathogen are studied in their natural hosts – typically support substantial virus replication, antiviral inflammatory responses and more severe disease. Of these cognate models, bovine RSV is the most closely related to the human RSV pathogen in terms of direct amino acid sequence homology, although cattle infection models are expensive and require extensive veterinary expertise for appropriate experimental use. The mouse pneumovirus, pneumonia virus of mice (PVM), undergoes robust replication in mouse bronchial epithelial cells in vivo and replicates the clinical symptoms of the more severe forms of human RSV disease in most inbred strains of mice (reviewed in ). With this pathogen, one can take advantage of the tools available for working in mouse models; however PVM is antigenically distinct from human RSV, and special arrangements need to be made to prevent cross-contamination when working in a general mouse vivarium.
Inflammatory responses and experiments designed to examine therapeutic approaches in these model systems will be presented in the sections to follow.
While the events that comprise the inflammatory response are multi-faceted and complex, the definition of inflammation, from the perspective of pathophysiology, is quite straightforward. From the Mondofacto Online Medical Dictionary : Inflammation: a localized protective response elicited by injury or destruction of tissues, which serves to destroy, dilute or sequester both the injurious agent and the injured tissue. Thus, inflammation elicited in response to hRSV is a series of events designed to destroy, dilute and/or sequester the virus and the tissue that is injured, either as a direct result of the virus replication, or as “collateral damage.” Indeed, much of the difficulty and damage that ensues from uncontrolled inflammation, also known as the “cytokine storm” emerges from this point, i.e., the inability to limit and sequester damaged tissue and to proceed onwards to a resolution phase.
Two recent studies provided valuable and dramatic insight into the acute inflammatory responses to human RSV infection. Johnson and colleagues  presented a thorough pathologic examination of lung tissue from an infant who died of acute trauma shortly after an outpatient diagnosis of RSV; this analysis was accompanied by post-hoc evaluation and immunohistochemical diagnosis of archived tissue samples from patients dating back as early as 1925. In an independent study, Welliver and colleagues  presented an analysis of lung tissue from eleven recent deaths due to acute RSV infection. The differences between the reports were as striking as the similarities. While both groups detected prominent RSV antigen in lung tissue, Johnson and colleagues detected antigen in bronchiolar epithelial cells in the tissue sample from the acute case; RSV antigen was detected in alveolar debris in one of the archival samples. Welliver and colleagues reported dramatic levels of RSV antigen in exfoliated alveolar cells only, as the fatal cases in their group had likely progressed much further toward tissue damage and pathology. More striking, Johnson and colleagues detected neutrophils predominantly in the airway lumen, mononuclear cells (including CD8+T cells) within the lung parenchyma, and prominent peribronchial/periarterial lymphoid (B-cell) aggregates. In contrast, the tissues evaluated by Welliver and colleagues were nearly devoid of T cells (CD4+ or CD8+), were likewise devoid of NK cells, but had abundant neutrophils and macrophages throughout. While it is difficult to classify the archived samples in the Johnson and colleagues study unequivocally, certainly the simplest explanation of these differences is that the inflammatory picture observed in the child who died of acute trauma was that of an early RSV infection, in the earliest phases of control or resolution; the latter study, by Welliver and colleagues presents the acute inflammatory sequelae of fatal, uncontrolled inflammatory disease. The molecular basis of these differences will emerge as we begin to understand more about the inflammatory pathways and immunomodulatory control of this illness.
There are numerous studies reporting findings on cells and mediators detected in bronchoalveolar lavage (BAL) fluid, typically from mechanically-ventilated infants and children, and likewise, nasopharyngeal washings, in association with acute RSV infection. The highlights will be featured here, with the intent of pointing to some general trends. The neutrophil chemoattractant interleukin (IL)-8, the proinflammatory cytokine, TNFα, and the CC chemokines MIP-1α, MCP-1 and RANTES are among those that are detected most frequently in the airways and nasal passages of infected infants and children [28–32]; likewise, varying levels of Th1 (interferon (IFN)γ) and Th2 (IL-4, IL-5) cytokines have been reported, the latter group typically elevated in the youngest RSV-infected infants [33–35]. The predominant Th2 response in the younger infants has been associated with a predilection to go on to develop childhood wheezing [36, 37], but cause and effect has not been definitively established. Several of the aforementioned findings include notable disease-related correlations; for instance, Garofalo and colleagues  documented a correlation between levels of MIP-1α in nasopharyngeal secretions and disease severity; likewise, several groups have documented correlations between chemoattractants and levels of eosinophil granule proteins in lung tissue and serum [39–42]. Mukherjee and colleagues  recently reported detection of the Th17 cytokine, IL-17 in tracheal aspirates from RSV-infected intubated infants. It is important to recognize that there remains no consensus on the relative importance of any specific cytokine or chemokine toward disease severity or amelioration, nor is at all clear whether it will be possible to target a unique cytokine or signaling pathway, once the interplay and interactions between them have set off a multi-ricocheting “cytokine storm”. Nonetheless, individual pathways have been targeted in several preliminary model experiments.
Similarly, several cytokine and related innate immune gene polymorphisms have been associated with RSV disease severity, including those in the human gene encoding IL-8, IL-4 and the IL-4 receptor α, IL-10, IL-18, RANTES, and surfactant proteins SP-A and SP-D [44 – 49]. As with the cytokine levels measured in BAL and nasopharyngeal fluids, these discoveries are most intriguing, but it is not yet clear how any one gene polymorphism informs clinical management or contributes to disease pathogenesis.
Bovine RSV (bRSV) is a significant pathogen of newborn calves; virus replication and seasonal disease patterns are similar to that observed in humans. As this experimental model has been explored primarily as a target for bovine vaccine technology, there is somewhat limited information on the specifics of this cognate virus-host inflammatory response. Gershwin and colleagues  described the results of a model of experimental bRSV infection, noting that the infected calves had varying degrees of necrotizing and proliferative bronchiolitis and alveolitis with syncytial formation, with some progressing to emphysematous bullae in the caudal lung lobes. Valarcher and Taylor  likewise described experimental infection of gnotobiotic (germ-free) calves resulting in increased levels of transcripts encoding IL-12, IFNγ, TNF-α, IL-6 IL-18, IL-8, RANTES, MCP-1 and MIP-1α, similar to findings reported in human subjects.
Ackermann and colleagues [52, 53] recently introduced the preterm lamb model, in which lambs delivered 9 days prior to full term are inoculated with human or bovine RSV. These lambs respond with a heightened inflammatory response (MCP-1, MIP-1α, IFNγ, TNF-α) in association with neutrophil recruitment, elevated levels of viral antigen, and overall increased degree of disease severity. This is an intriguing new model; ongoing explorations include roles of surfactant protein and responses to vascular endothelial growth factor, discussed further in sections to follow.
Inbred BALB/c mice have been used extensively to explore the inflammatory responses to the human RSV pathogen (reviewed in ). Graham and colleagues  published the first examples of histopathology of susceptible BALB/c mice challenged with high titer inocula of hRSV, which included perivascular and peribronchial infiltrates, as well as lymphocytes and macrophages in the alveolar spaces, most severe between days 5 – 8, resolving by day 10 after inoculation. Subsequent studies identified numerous proinflammatory mediators produced and released in association with hRSV challenge, including MCP-1, RANTES, the IFNγ regulated protein, IP-10, KC, and MIP-1α [55–58], and more recently, IL-17 [43, 59]. Other important findings established with the RSV challenge model include the predilection for a Th2 skewed secondary response in neonatal mice [60–62], the role of toll-like receptors in receiving and transducing proinflammatory signals (reviewed in ) and the interplay of macrophages, T lymphocytes (including regulatory T cells), dendritic cells and NK cells in modulating the inflammatory state [64–67].
As noted above, cotton rats have been developed as an alternative infection model for preclinical vaccine research. In an early report on this model, Prince and colleagues  presented lung tissue from a 6-week-old cotton rat, documenting a relatively mild proliferative bronchiolitis, most acute on days 5–6 after inoculation with hRSV. Bronchial lavage of hRSV inoculated cotton rats results in recovery of primarily activated macrophages and lymphocytes  similar to what has been described for the hRSV-BALB/c mouse challenge. In subsequent publications, Blanco and colleagues  and Boukhvalova and colleagues [71, 72] reported expression of transcripts encoding cytokine and chemokine genes in lung tissue isolated from cotton rats inoculated with hRSV, including RANTES, IP-10, MCP-1, IL-6, MIP-1β, IFNγ and TNF-α and also the role of TLR-3 responses in promoting an antiviral pathology.
PVM is a natural rodent pneumovirus pathogen and elicits severe inflammatory pathology in most inbred strains of mice . Virus replication can initially be detected in bronchial epithelial cells; findings at peak morbidity include high virus titer, prominent neutrophil influx, and edema , similar to that described by Welliver and colleagues for fatal RSV infection . Cytokines detected in association with acute PVM infection also include MIP-1α, IP-10, MCP-1, IFNγ, and TNF-α.
As is clear from the human pathology studies, from the findings from the airways of mechanically ventilated infants, and from various infection models, the most severe forms of pneumovirus infection are associated with significant inflammatory pathology. What remains at issue – can severe disease be averted by administration of antiviral agents alone? In order to answer this question, one needs to consider first the case of ribavirin, which was previously administered routinely to all infants hospitalized with RSV disease. Although ribavirin was quite effective as an antiviral, the clinical impact was determined to be minimal [7, 8]. We have examined this phenomenon with PVM, an infection model that includes robust virus replication accompanied by a prominent inflammatory response, which leads to significant morbidity and mortality in the absence of intervention. Ribavirin has antiviral activity against PVM; administration of 37.5 mg/kg twice per day resulted in complete cessation of viral replication within hours of the first dose. We chose day 3 after PVM inoculation as the point to initiate ribavirin therapy, as this was the first day on which the mice appeared to be “symptomatic” (as in, slightly ruffled fur); this situation would replicate as closely as possible what might occur in a clinical setting. We found that, although ribavirin was remarkably effective at putting a stop to virus replication, it had virtually no impact on the production of proinflammatory mediators CCL2 and CCL3, no effect on the ongoing recruitment of inflammatory cells, nor any impact on disease-associated mortality . Thus, we concluded that, by day 3 of PVM infection, the course and clinical outcome of the disease process was not defined by virus replication alone. We hypothesized that one or more components of the inflammatory response had been initiated and was amplified in a manner that was independent of ongoing virus replication, and as such, was not controlled even by a strong and effective antiviral, such as ribavirin.
There are a number of interesting and novel antiviral approaches to RSV described in the literature, including antisense targeting mechanisms that have enhanced antiviral potency and may have additive or even synergistic impact with ribavirin [75–77]. But the question remains: if one cannot predict infection before the inflammatory response has taken off on an independent trajectory, will any antiviral alone ever be enough?
Following through on the experiment described above, we found that when we added cytokine signaling blockade – specifically by blocking the actions of the proinflammatory cytokine, CCL3 (MIP-1α) – we were able to augment survival significantly, from 20% in response to ribavirin alone to 60% in response to ribavirin together with anti-MIP-1α monoclonal antibody; (Figure 1; ). Survival in response to acute PVM infection was also enhanced in response to administration of metRANTES, which blocks CCL3 signaling via its receptor, CCR1, expressed on leukocytes . In a similar experiment performed by Culley and colleagues , metRANTES was administered to BALB/c mice that received a primary intranasal inoculum of RSV A2, resulting in diminished expression of CCL5 (RANTES) transcript immediately after inoculation, reduced virus clearance, and diminished recruitment of CD4+ and CD8+ T cells. Likewise, Miller and colleagues  demonstrated diminished airways hyper-responsiveness and reduced mucus production in association with increased levels of IFNγ and CXCL10 (IP-10) in RSV-challenged mice devoid of CCR1.
A parallel example of cytokine signaling blockade was reported by Lukacs and colleagues . In this work, the authors explored the role of CCR6, a chemokine receptor on T cells and dendritic cells, on the outcome of RSV challenge, also in the BALB/c mouse model. Among the findings, CCR6 gene-deleted mice expressed diminished levels of Th2 cytokines (IL-4, IL-5 and IL-13), diminished airways hyper-reactivity, and diminished virus recovery at day 3 post-inoculation.
Another interesting approach was taken by Hancock and colleagues . Given the role of CCR5 as a co-receptor for human immunodeficiency virus (HIV), the authors determined that treatment of target HEp-2 cells with recombinant human RANTES resulted in dose-dependent resistance to RSV infection, an observation that was unique to this chemokine and not shared with any of a large group of similar proteins. While these findings were intriguing, the fact that epithelial cells do not ordinarily express CCR5 provides a curious puzzle to solve. As RANTES is produced prominently by macrophages  in response to mouse RSV challenge, perhaps it has some unique immunomodulatory impact on target cells.
The findings of Casola and colleagues , focused on reducing oxidative stress in association with RSV-induced lung inflammation. In previous work, the authors demonstrated that RSV regulates both reactive oxygen and CCL5 (RANTES) production . The anti-oxidant butylated hydroxyanisole (BHA) was administered orally to BALB/c mice challenged with hRSV. Weight loss, lipid peroxidation products, neutrophil recruitment, cytokine production, and airways hyperreactivity were all reported as diminished in response to treatment, although interesting, BHA had no impact on virus recovery.
Leukotrienes are prominent among the inflammatory mediators produced in response to RSV. Welliver and colleagues  demonstrated that systemic administration of the 5-lipoxygenase inhibitor, zileuton, led to diminished leukotriene synthesis, reduced inflammation (measured as number of lymphocytes and macrophages in the airways), and diminished airway obstruction in response to RSV inoculation in the BALB/c mouse model.
The cysteinyl-leukotriene inhibitor, montelukast was utilized to explore the role of leukotriene antagonism in the PVM model . Interestingly, montelukast monotherapy, initiated on day 3 after virus inoculation, had no substantial impact on limiting granulocyte recruitment to the airways, nor did it improve clinical outcome from acute virus infection. However, montelukast together with ribavirin resulted in diminished mortality. Montelukast was also featured in a recent study of post-infectious airway dysfunction in the BALB/c RSV challenge model. Han and colleagues  treated mice with montelukast prior to RSV challenge and for several days thereafter, a strategy that prevented AHR, airway eosinophilia, and mucus hyperproduction upon re-challenge with RSV.
Montelukast was also highlighted in a successful clinical trial aimed at prevention of post-RSV reactive airway disease . Infants hospitalized with RSV bronchiolitis were randomized to treatment and non-treatment groups, the treatment group receiving montelukast tablets (and the non-treatment, matching placebo) within 7 days of RSV symptoms emerging. The study group reached the conclusion that montelukast treatment as described effectively reduced numerous negative lung inflammatory sequelae subsequent to RSV bronchiolitis.
While the anti-sense oligonucleotides have been used primarily as a means to target nascent virus transcripts, Cormier and colleagues  have expanded on this approach and used this method to target the interleukin-4 receptor α chain during primary RSV challenge in neonatal mice. Anti-sense oligonucleotides were administered intranasally in conjunction with hRSV challenge, which resulted in an increased Th1 cytokine response in these younger mice, and prevented the characteristic pulmonary dysfunction observed on re-challenge that has been attributed to the Th2 skewing in the neonates. This immunomodulatory approach - both the target and the methodology - has broad implications for the management of neonatal RSV infection.
Pulmonary surfactant is the phospholipoprotein formed by type II alveolar cells, and typically demarcates lung maturity. Surfactant proteins A and D (Sp-A and Sp-D), also known as collectins, have become the subject of studies focused on antiviral and anti-inflammatory host defense . LeVine and colleagues  examined RSV challenge in Sp-A gene-deleted mice, and found augmented neutrophil recruitment, elevated levels of TNFα and IL-6 in lung tissue in association with elevated virus titers. This work sets the background for the recent findings of Numata and colleagues , who reported on the therapeutic efficacy of palmitoyl oleoyl-phosphatidyl glycerol (POPG), a minor but apparently important component of pulmonary surfactant, which blunted the global inflammatory response when administered together with hRSV. Mechanistically, POPG interacted with the TLR4-associated proteins CD14 and MD2 in cell culture studies, which resulted in diminished production of both IL-6 and IL-8. In vivo, POPG administration resulted in reduced recovery of infectious virus and elimination of the IFNγ response. It is not yet clear whether POPG would be equally effective when administered either as a therapeutic agent, or as a prophylactic agent at time points preceding virus inoculation. While surfactant-supplementation and replacement strategies have been attempted for the treatment of infants with severe RSV bronchiolitis [93, 94], the value of this therapy remains as yet unproven .
Meyerholz and colleagues  approached this theme from another perspective, and treated newborn lambs with recombinant human vascular endothelial growth factor (VEGF), one of the factors known to regulate expression of surfactant proteins. Lambs that received intratracheal VEGF prior to inoculation with hRSV responded with diminished recruitment of neutrophils to the airways, but interestingly, no change in expression of surfactant proteins A or D. The molecular basis for these observations remains to be clarified.
Clearly, the inflammatory responses to pneumovirus infection are multi-faceted, complex, and refractory to standard approaches, such as antivirals and glucocorticoids. There has been great interest in the literature regarding the immumodulatory impact of probiotic/pharmabiotic microorganisms, which alter the responses of mucosal tissue to subsequent proinflammatory challenge (reviewed in ). Lactobacilli have been particularly useful in reducing gastrointestinal inflammation when administered orally. We have found that live lactobacilli, when targeted to the respiratory epithelium, are highly effective at suppressing pneumovirus-induced inflammation and provided full (100%) protection against lethal disease even without the addition of a replication inhibitor . Most interesting, this protection was prolonged; priming with live lactobacilli can take place 3 – 5 months before the virus inoculation, yet significant protection persists. Protection was associated with diminished granulocyte recruitment and global suppression of the virus-induced “cytokine storm” [Figure 2] which remained effective in mice devoid of the TLR signaling adapter, MyD88.
Human RSV is an important respiratory infection that has an impact on infants, children and the elderly. While there is effective antibody prophylaxis for identified high-risk cases, there is no specific therapy for severe disease, nor is there a vaccine. Our findings and those of others suggest that replication inhibition alone may not be effective at preventing the negative sequelae of severe infection. Specifically – once an individual experiences symptoms, the inflammation may become uncontrolled and no longer linked directly to virus replication; as such, the impact of replication inhibitors will be negligible. Immunomodulatory measures may ultimately lead to more effective control of severe respiratory syncytial virus disease.
Dr. Rosenberg’s laboratory is supported by NIAID Division of Intramural Research (AI000943). Dr. Domachowske’s laboratory is supported by the Children’s Miracle Network of New York.